Mesenteric/celiac duplex ultrasound interpretation criteria revisited

Transcription

1 From the Southern Association for Vascular Surgery Mesenteric/celiac duplex ultrasound interpretation criteria revisited Ali F. AbuRahma, MD, a Patrick A. Stone, MD, a Mohit Srivastava, MD, a L. Scott Dean, PhD, MBA, b Tammi Keiffer, RN, b Stephen M. Hass, MD, a and Albeir Y. Mousa, MD, a Charleston, WV Background: Several published studies with a small sample size have reported differing results of duplex ultrasound (DUS) utilizing different threshold velocities in detecting significant stenosis of superior mesenteric () or celiac arteries (CA). The present study is based on the largest number of mesenteric duplex/angiography correlations reported to date for the diagnosis of /CA stenosis. Methods: One hundred fifty-three patients (151 and 150 CA) had both DUS and arteriography. Receiver operator curves (ROC) were used to analyze peak systolic velocity (PSV), end diastolic velocity (EDV), and or CA/aortic PSV ratio in detecting >50% and >70% stenosis. Results: For (151 arteries: 84 with >50% stenosis [54 of which had >70% stenosis] based on angiography): the PSV threshold that provided the highest overall accuracy (OA) for detecting >50% stenosis was >295 cm/s (sensitivity [sens.] 87%, specificity [spec.] 89%, and OA 88%); and for detecting >70%, it was >400 cm/s (sens. 72%, spec. 93%, and OA 85%). The EDV threshold that provided the highest OA for detecting >50% stenosis was >45 cm/s (sens. 79%, spec. 79%, and OA 79%); and for >70% stenosis was >70 cm/s (sens. 65%, spec. 95%, and OA 84%). ROC analysis showed that PSV was better than EDV and /aortic PSV ratio for >50% stenosis of (P.003 and P.0005). For celiac arteries (150 arteries: 105 with >50% stenosis [62 of which had >70% stenosis]): the PSV threshold that provided the highest OA for >50% stenosis was >240 cm/s (sens. 87, spec. 83%, and OA 86%); and for >70% stenosis was >320 cm/s (sens. 80%, spec. 89%, and OA 85%). The EDV threshold that provided the highest OA for >50% stenosis was >40 cm/s (sens. 84%, spec. 48%, and OA 73%); and for >70% stenosis was >100 cm/s (sens. 58%, spec. 91%, and OA 77%). ROC analysis showed that PSV was better than EDV and /aortic PSV ratio for >50% stenosis of CA (P <.0001 and P.0410.) Conclusions: PSV values can be used in detecting >50% and >70% /CA stenosis and were better than EDVs and ratios. Previously published data must be validated in individual vascular laboratories. Our results will need prospective validation. (J Vasc Surg 2012;55: ) Since Nicholls and Strandness reported on the use of hemodynamic parameters in the diagnosis of mesenteric insufficiency, 1 several published reports with a small sample size have reported conflicting results of duplex ultrasound utilizing different threshold velocities in detecting significant ( 50% and 70%) stenosis of the superior mesenteric () or celiac arteries (CA). 2-7 Presently, there is no absolute consensus on specific duplex ultrasound (DUS) criteria for the diagnosis of mesenteric/celiac artery stenosis. The present study is based on the largest number of mesenteric duplex/angiography correlations reported to date for the diagnosis of /CA stenosis and validates the optimal duplex velocity with the best positive predictive value (PPV) and negative predictive value (NPV) in detecting 50% to 70% stenosis of the and CA. From the Department of Surgery, Robert C. Byrd Health Sciences Center, West Virginia University, a and Charleston Area Medical Center. b Competition of interest: none. Presented at the Thirty-fifth Annual Meeting of the Southern Association for Vascular Surgery, Naples, Fla, January 19-22, Additional material for this article may be found online at Reprint requests: Ali F. AbuRahma, MD, Department of Surgery, Robert C. Byrd Health Sciences Center, West Virginia University, 3110 MacCorkle Ave, SE, Charleston, WV ( ali.aburahma@camc.org). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a competition of interest /$36.00 Copyright 2012 by the Society for Vascular Surgery. doi: /j.jvs PATIENTS AND METHODS One hundred fifty-three patients (151 and 150 CA), in whom mesenteric ischemia was suspected and who had both DUS and mesenteric arteriography during a recent 8-year period were analyzed. A list of patients who underwent mesenteric angiography for chronic mesenteric ischemia was generated (patients with acute mesenteric ischemia were excluded), and this list was cross-referenced with the vascular laboratory report archive, which allowed identification of all patients who had both DUS and angiography within 30 days. Only patients with technically satisfactory mesenteric DUS of at least the or CA and angiography were analyzed. Thirteen patients were excluded from analysis, six of whom had extensively calcified plaques at the origin of the or CA on angiography, which made exact calculation of stenosis difficult, and seven others who had inadequate duplex examinations because of excessive gas, which did not allow satisfactory Doppler velocity sampling of the and CA. In two patients, the was not visualized/recorded and in three others, the CA was not visualized/recorded, but the other vessels were seen on DUS, and both were included in this analysis. This study was approved by the Institutional Review Board of Charleston Area Medical Center/West Virginia University, Charleston Division. Mesenteric duplex ultrasound technique. All testing was done by registered vascular technologists in our accredited vascular laboratory (Intersocietal Commission for the

2 JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 AbuRahma et al 429 Table I. Mean PSVs, EDVs, and /aortic PSV ratios for various categories of angiographic stenosis Label Mean Median Standard deviation Minimum Maximum P value (PSV) Normal a (EDV) Normal Systolic ratio Normal EDV, End diastolic velocity; PSV, peak systolic velocity;, superior mesenteric artery. a Compares the velocities across the stenosis levels. Accreditation of Vascular Laboratories) using a Phillips system (ATL, 5000 instrumentation; ATL Ultrasound, Bothell, Wash) and utilizing low-frequency 2- to 5-MHz curvilinear phased array transducers. Every effort was made to utilize a Doppler angle of 60 degrees or less to provide consistency in the Doppler velocities measurement. If visualization of the mesenteric vessels was limited in the supine position because of overlying bowel gas, the oblique position was used (which was successful in one patient) to better visualize these vessels using the liver as an acoustic window. The examination was started by evaluating the aorta by placing the transducer just below the xiphoid process. Transverse and sagittal planes were used for visualization. Attention was then turned to the, which was best seen on the sagittal view arising from the anterior aspect of the aorta. Since the has a distinctive anatomic location, it serves as a landmark for scanning other mesenteric vessels in the upper abdomen. The can also be seen in the transverse view since it is surrounded by a prominent ring of retroperitoneal fat that separates the from the pancreas. 8 The CA origin is generally best accomplished in the sagittal plane, whereas its main branches (common hepatic, splenic, and left gastric arteries) are best seen in the transverse view. Its classical ultrasound visualization of the T- shaped bifurcation (seagull sign) on the transverse view is a characteristic landmark. Assessment of the gray-scale and color Doppler evaluation of the aorta and the mesenteric vessels is accomplished by searching for the presence of atherosclerotic plaques and luminal narrowing. This is followed by color Doppler analysis. The sample volume (1.5 mm) must be passed slowly from the aorta into the orifice and the proximal segment of each vessel, searching for elevated peak systolic velocities (PSV) or end diastolic velocities (EDV) and poststenotic turbulence and bruits. All duplex scans were interpreted by Board certified vascular surgeons with the Registered Physicians Vascular Interpretation (RPVI) credentials. Arteriographic evaluation was performed using both abdominal aortography with lateral projection and selective intra-arterial digital subtraction of the mesenteric vessels using a SOS Omni catheter (AngioDynamics, Latham, NY). These were performed by four board-certified vascular surgeons and two board-certified vascular interventionalists. These physicians have extensive experience in both diagnostic and therapeutic endovascular procedures (with over 2000 cases annually). The points of maximal stenosis were measured using calipers and then divided by the diameter of the distal artery to calculate the presence of stenosis. The angiograms were interpreted by physicians (who did not perform the angiography) who were blinded to the DUS findings. Statistical analysis. The data analysis was performed using SAS 9.2 (SAS Institute Inc, Cary, NC) and Sigma Plot 10 (Systat Software, Inc, Chicago, Ill). The velocity data were expressed as a mean plus or minus standard deviation. Analysis of variance (ANOVA) test was used to compare the means of PSV, EDV, and celiac/aortic and /aortic ratio to the stenosis categories of, 30% to 50%, 50% to 69%, and 70% to 99%. We performed two types of receiver operator curves (ROC) analyses comparing angiographic measurements of stenosis with duplex velocity measurements to determine the optimum velocity criteria. In the primary analysis we included all patients with interpretable angiographic images and duplex scans, including patients with angiographic occlusions. This allowed us to analyze the following categories of angiographic stenosis in comparison to velocity measurements obtained with duplex scanning: 30% to 100%, 50% to 100%, and 70% to 100%. A second analysis excluded patients with angiographic occlusions. In this analysis, we analyzed the following categories of angiographic stenosis in comparison to velocity measurements obtained with duplex scanning: 30% to 99%, 50% to 99%, and 70% to 99%. The method of DeLong, DeLong, and Clarke-Pearson 9 was used to compare ROC areas for the paired data. The differ-

3 430 AbuRahma et al JOURNAL OF VASCULAR SURGERY February 2012 Table II. Sensitivity, specificity, PPV, NPV, and OA for the best velocity cutoffs and ratios for various categories of angiographic stenosis Group No flow or value Sensitivity (CI) Specificity (CI) PPV NPV Accuracy Occlusion included in analysis - 30% stenosis (PSV) ( ) 94 ( ) (EDV) ( ) 71 ( ) /aortic PSV ratios 2 93 ( ) 42 ( ) % stenosis (PSV) ( ) 89 ( ) (EDV) ( ) 79 ( ) /aortic PSV ratios ( ) 78 ( ) % stenosis (PSV) ( ) 93 ( ) ( ) 95 ( ) (EDV) ( ) 95 ( ) /aortic PSV ratios ( ) 83 ( ) Occlusion excluded from analysis - 30% stenosis (PSV) ( ) 94 ( ) (EDV) ( ) 76 ( ) /aortic PSV ratios 2 92 ( ) 42 ( ) % stenosis (PSV) ( ) 89 ( ) (EDV) ( ) 83 ( ) /aortic PSV ratios ( ) 78 ( ) % stenosis (PSV) ( ) 93 ( ) ( ) 95 ( ) (EDV) ( ) 96 ( ) /aortic PSV ratios ( ) 83 ( ) EDV, End diastolic velocity; OA, overall accuracy; PSV, peak systolic velocity;, superior mesenteric artery. ence of each area pair and its standard error and 95% confidence interval were computed. This was followed by the 2 statistic for the area comparison and its associated P value. The sensitivity, specificity, PPV, NPV, and overall accuracy (OA) were determined for specific PSV and EDV values and celiac/ aortic and /aortic PSV ratios. /CA anatomic anomalies were included in the statistical analysis irrespective of the angiographic anomaly. A significance level of 0.05 was used to determine statistical significance. RESULTS The mean age of the group was 67.8 years (range: years). Seventy-three percent were females and 27% were males.. This study included 151 s, which included 35 (23%) ( 30%), 32 (21%) with 30 to 50% stenosis, 30 (20%) with 50% to 69% stenosis, and 54 (36%) with 70% stenosis (including seven with occlusion) based on angiography. Two s were not visualized/recorded on duplex ultrasound. Table I summarizes the mean PSVs, EDVs, and ratios for various categories of stenosis. The mean PSV, EDV, and ratios were statistically significantly different between, 50%, 50% to 69%, and 70% to 99% stenosis (P.0001). Table II summarizes the sensitivity, specificity, PPV, NPV, and OA for the best velocity cutoffs and ratios for the diagnosis of 30%, 50%, and 70% stenosis of the. As noted, a PSV of 210 cm/s had the best accuracy (93%) for detecting 30% stenosis with a sensitivity of 93% and a specificity of 94%. Meanwhile, a PSV of 295 cm/s had the best OA (88%) in detecting 50% stenosis, with a sensitivity of 87% and a specificity of 89%. A PSV of 400 cm/s had the best OA of 85%, with a sensitivity of 72%, and a specificity of 93% in detecting 70% stenosis. If occluded s (based on angiography) were excluded (Appendix Tables I, A-C, online only), the same cutoff values exist, while the OA is slightly lower for the analysis that excluded occlusions. Appendix Tables I, A-C (online only) are included in the online appendix and they summarize various velocity cutoffs and ratios for the diagnosis of 30%, 50%, and 70% stenosis. Overall, there were two patients with 99% stenosis of the and seven patients with occlusion, based on angiography. In the two patients with 99% stenosis, one patient had a PSV of 607 cm/s and EDV of 92 cm/s, and the other one had a PSV of 145 cm/s and an EDV of 28 cm/s. Of the seven patients with occluded s, six were detected with DUS demonstrating a muffled monophasic low velocity distal to calcification, which was seen on B- mode imaging.

4 JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 AbuRahma et al 431 Fig 1. Receiver operator curves comparing the peak systolic velocities (PSVs), end diastolic velocities (EDVs), and superior mesenteric artery ()/aortic PSV ratios for 50% stenosis. AUC, Area under the curve. Fig 2. Receiver operator curves comparing the peak systolic velocities (PSVs), end diastolic velocities (EDVs), and superior mesenteric artery ()/aortic PSV ratios for 70% stenosis. AUC, Area under the curve. ROC analysis of velocities. Figs 1 and 2 are ROC curves plotting sensitivity against specificity for diagnosing 50% and 70% stenosis. As noted in these figures, the PSV was better than the EDV and ratio in detecting 50% stenosis (P.003 and P.0005, respectively, Fig 1). Meanwhile, the PSV was superior to the ratio in detecting 70% (P.0237), and the EDV was also significantly better than the ratio (P.0485 Fig 2). Celiac artery. One hundred fifty CA were analyzed, including 24 (16%) ( 30%), 21 (14%) with 30 to 50% stenosis, 43 (28%) with 50% to 69% stenosis, and 62 (41%) with 70% stenosis based on angiography. Three out of 153 CA were not visualized or recorded on DUS. Table III summarizes the mean PSVs, EDVs, and ratios for various categories of CA stenosis. As noted, the mean PSV, EDV, and ratios were statistically significantly different between, 50%, 50% to 69%, and 70 to 99% stenosis (P.0001). Table IV summarizes the sensitivity, specificity, PPV, NPV, and OA for the best velocity cutoffs and ratios for the diagnosis of 30%, 50%, and 70% stenosis of the CA. As noted, a PSV 180 cm/s had an OA of 96% with a sensitivity and specificity of 96% and 95%, respectively in detecting 30% stenosis. The PSV threshold that provided the highest OA for 50% stenosis of the CA was 240 cm/s (sensitivity [sens.] 87%, specificity [spec.] 83%, and OA 86%); and for 70% stenosis was 320 cm/s (sens. 80%, spec. 89%, and OA 85%). If CA occlusions were excluded from analysis, the same cutoff values exist, while the OA is slightly lower for the analysis that excluded occlusions. Appendix Tables II, A-C (online only) are included in the online appendix and summarize the various velocity cutoffs and ratios for the diagnosis of 30%, 50%, and 70% CA stenosis. Overall, one patient was diagnosed with 99% stenosis of the CA and four with CA occlusion on angiography. Three of these occlusions were detected on DUS, and the remaining CA occlusion had low velocities (a PSV of 52 cm/s and an EDV of 18 cm/s). Meanwhile, the patient with 99% stenosis had a PSV of 238 cm/s and an EDV of 45 cm/s. ROC analysis of CA velocities. Figs 3 and 4 are ROC curves plotting sensitivity against specificity for diagnosing 50% and 70% CA stenosis. As noted in these figures, the PSV was better than the EDV and PSV ratio in detecting 50% stenosis (P.0001 and P.0410, respectively, Fig 3). Meanwhile, the PSV was superior to the EDV in detecting 70% (P.0258, Fig 4). /celiac artery anatomic anomalies. Overall, seven patients in this series had anatomic anomalies of the /CA on arteriography; five of them had right hepatic artery arising from the and two patients had a common origin of the celiac and. DISCUSSION Sonographic mesenteric vessel imaging has remained a vital tool in deciding whether to proceed with invasive testing, however, to date, there have been only a few series in the literature describing duplex velocity criteria in detect-

5 432 AbuRahma et al JOURNAL OF VASCULAR SURGERY February 2012 Table III. Mean PSVs, EDVs, and celiac artery/aortic PSV ratios for various categories of angiographic celiac artery stenosis Label Mean Median Standard deviation Minimum Maximum P value Celiac (PSV) Normal a Celiac (EDV) Normal Systolic ratio Normal EDV, End diastolic velocity; PSV, peak systolic velocity;, superior mesenteric artery. a Compares the velocities across the stenosis levels. Table IV. Sensitivity, specificity, PPV, NPV, and OA for the best velocity cutoffs and ratios for various categories of angiographic celiac stenosis Group No flow or value Sensitivity (CI) Specificity (CI) PPV NPV Accuracy Occlusion included in analysis Celiac - 30% stenosis Celiac (PSV) ( ) 95 ( ) ( ) 100 ( ) Celiac (EDV) ( ) 5 ( ) Celiac/aortic PSV ratios ( ) 20 ( ) Celiac - 50% stenosis Celiac (PSV) ( ) 83 ( ) Celiac (EDV) ( ) 48 ( ) ( ) 58 ( ) Celiac/aortic PSV ratios ( ) 71 ( ) Celiac - 70% stenosis Celiac (PSV) ( ) 89 ( ) Celiac (EDV) ( ) 91 ( ) ( ) 92 ( ) ( ) 95 ( ) Celiac/aortic PSV ratios ( ) 87 ( ) Occlusion excluded from analysis Celiac - 30% stenosis Celiac (PSV) ( ) 95 ( ) ( ) 100 ( ) Celiac (EDV) ( ) 5 ( ) Celiac/aortic PSV ratios ( ) 20 ( ) Celiac - 50% stenosis Celiac (PSV) ( ) 83 ( ) Celiac (EDV) ( ) 48 ( ) ( ) 58 ( ) Celiac/aortic PSV ratios ( ) 71 ( ) Celiac - 70% stenosis Celiac (PSV) ( ) 89 ( ) Celiac (EDV) ( ) 91 ( ) ( ) 92 ( ) ( ) 95 ( ) Celiac/aortic PSV ratios ( ) 87 ( ) EDV, End diastolic velocity; NPV, negative predictive value; OA, overall accuracy; PPV, positive predictive value; PSV, peak systolic velocity.

6 JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 AbuRahma et al 433 Fig 3. Receiver operator curves comparing the peak systolic velocities (PSVs), end diastolic velocities (EDVs), and celiac artery/aortic PSV ratios for 50% celiac artery stenosis. AUC, Area under the curve. Fig 4. Receiver operator curves comparing the peak systolic velocities (PSVs), end diastolic velocities (EDVs), and celiac artery/aortic PSV ratios for 70% celiac artery stenosis. AUC, Area under the curve. ing significant stenosis of the mesenteric vessels. In addition, each of these series had a small sample size and they offered differing evidence on the best parameter to use when defining significant mesenteric stenosis. 2-4,5,7 In 1991, Moneta et al published a hypothesis seeking a study for diagnosing mesenteric stenosis based on duplex scanning. This initial study analyzed only 26 arteries with 70% stenosis of the or CA. They analyzed several different variables and established that a PSV of 200 cm/s in the CA and 275 cm/s in the predicted stenosis of 70%. An EDV of 55 cm/s in either artery produced similar sensitivities and specificities, although it was not as accurate as the PSV across the ostium of the artery. All of the other ratios ( or CA/aortic ratios) that were examined did not improve on either of these two variables. 3 In 1993, the same group reported on a larger blinded prospective series involving 100 patients who underwent routine mesenteric duplex scanning and lateral abdominal aortography, regardless of their abdominal symptoms. Twenty-three of these patients were found to have 70% stenosis of the CA and 13 had 70% stenosis of the. Using the same criteria, a sensitivity, specificity, and OA of 92%, 96%, and 96% were attained for diagnosing 70% stenosis in the. For the CA, the results were not as good, but showed a sensitivity and specificity of 87% and 80%, respectively, along with an OA of 82%. This validated the group s previously published data, along with reestablishing the PSV as the single best parameter for diagnosing the percentage of stenosis in both the and CA. 4 In 1991, Bowersox et al found that the EDV was actually a better variable in their laboratory for detecting the percentage of stenosis. In their study of 24 patients, an EDV of 45 cm/s was the best value for detecting 50% stenosis of the, with a sensitivity and specificity of 100% and 92%, respectively. 2 Zwolak et al, from the same group, confirmed similar findings in a series of 46 visceral arteries, where an EDV of 45 cm/s yielded a sensitivity and specificity of 90% and 91%. Increasing the threshold to 55 cm/s for the CA showed a similar sensitivity and specificity, with an OA of 95%. 5 When Moneta et al 4 applied the same EDV values ( 45 cm/s) to his own validation study in 1993, the sensitivity was significantly less at 48%, with a specificity of 99% and OA of 86%. Our present study is the largest to date to analyze each of these commonly used criteria. A total of 151 and 150 CA were visualized in this series, of which 84 and 105 CA were found to have 50% stenosis. Comparing previously published criteria may be difficult since some investigators sought to evaluate criteria for detecting 70% stenosis, 3,4 while others evaluated criteria for detecting 50% stenosis. 2,5,7 Therefore, our study analyzed different threshold velocities for detecting both 50% and 70% stenosis and compared them with previously published criteria. Our present study showed that the PSV of the or CA was the best variable for detecting both 50% and 70% stenosis. For, the most accurate PSVs for 50% and 70% stenosis were 295 cm/s and 400 cm/s,

7 434 AbuRahma et al JOURNAL OF VASCULAR SURGERY February 2012 Table V. A, Previously published mesenteric duplex criteria For stenosis: Using PSVs: Moneta criteria 3,4 : PSV of 275 cm/s for detecting 70% stenosis Bowersox criteria 2 : PSV of 300 cm/s for detecting 50% stenosis Using EDVs: Zwolak criteria 5 : EDV of 45 cm/s for detecting 50% stenosis Perko 7 : EDV of 70 cm/s for detecting 50% stenosis For CA stenosis: Using PSVs: Moneta criteria 3,4 : PSV of 200 cm/s for detecting 70% stenosis Zwolak criteria 5 : PSV of 200 cm/s for detecting 50% stenosis Using EDVs: Zwolak criteria 5 : EDV of 55 cm/s for detecting 50% stenosis Perko criteria 7 : EDV of 100 cm/s for 50% stenosis CA, Celiac artery; EDV, end diastolic velocity; PSV, peak systolic velocity;, superior mesenteric artery. respectively. These values showed a sensitivity, specificity, and OA of 87%, 89%, and 88% for 50% stenosis; and 72%, 93%, and 85% for 70% stenosis, respectively. Although our present study cannot be compared with previous studies (Table V, A) because of different designs, different equipment, and demographics, if Moneta s criteria (PSV of 275 cm/s for 70% stenosis) were used in our series, the sensitivity would be higher, at 92%, however, the specificity and OA would be lower. Applying Bowersox s use of a PSV 300 cm/s for 50% stenosis of the in our series would yield similar results with a slightly higher OA of 87%. When applying both Zwolak and Perko s criteria in our series, which utilized an EDV of 45 cm/s and 70 cm/s, respectively, in detecting 50% stenosis, it showed less sensitivity and specificity, along with an OA of only 79% and 68%, respectively (Table V, B). For the CA, a PSV of 240 cm/s and 320 cm/s showed the highest sensitivity, specificity, and OA for 50% and 70% stenosis at 87%, 83%, and 86% (for 50% stenosis) and 80%, 89%, and 85% (for 70% stenosis), respectively. If Moneta s criteria were applied (PSV of 200 cm/s for 70% stenosis), the sensitivity and NPV were higher at 98% and 97%, however, the specificity was only 35% with a PPV and OA of 52% and 61%, respectively. When Bowersox s criteria ( 200 cm/s for 50% stenosis) were applied, the sensitivity and OA were high at 95% and 85%, however, specificity was again lower at 60%. Similarly, when applying Zwolak s (EDV of 55 cm/s) and Perko s (EDV of 100 cm/s) criteria, both showed a relatively high specificity and PPV, but only produced an OA of 72% and 54%, respectively, in detecting 50% stenosis (Table V, C). The differing results of these previously published DUS criteria when applied to our series can be explained by a few possibilities: differences in instrumentation; Doppler angle; demographics, specifically gender, indications of the DUS (proportion of patients who presented with clinical manifestations of mesenteric ischemia); and the threshold of the angiographic stenoses ( 50% in contrast to 70%). Perhaps one of the most important reasons for these differences is the difference in instrumentation. It has been previously determined that Doppler velocity determination may vary among different instrumentation when evaluating the carotid bifurcation. 10 It should be noted that Moneta, et al 3,4 utilized an Acuson 128 color duplex scanner (Acuson, Inc, Mountain View, Calif), while Zwolak et al 5 utilized Diasonics equipment (DRF 400; Diasonics Ultrasound, Santa Clara, Calif) and Siemens instrumentation (Q 2000; Siemens Medical Systems, formerly Quantum Medical Systems, Issaquah, Wash) in their study, and we utilized a Phillips system (ATL 5000 instrumentation) in our present study. Another reason for the variations in these studies can be different Doppler angles. In our present study, every effort was made to keep the Doppler angle below 60 or 70 degrees to minimize angle-dependent variations in Doppler velocities. Some of the differences can also be gender-related. Seventy-three percent of our patients were females, in contrast to Moneta et al s study, which were primarily males, and Zwolak et al s study, which was primarily females. As noted in our ROC analysis, the systolic ratio was not as accurate as the PSV in detecting 50% and 70% stenosis and 50% CA stenosis. A similar observation was noted by other authorities. 4,5,7 Overall, seven patients in this series had anatomic anomalies of the /CA artery on arteriography; five of whom had right hepatic artery arising from the, four of these were confirmed by DUS and had PSVs consistent with 50% stenosis, and the remaining patient had a PSV consistent with 50% stenosis. The other two patients had a common origin of the CA and, and both were confirmed with DUS, and one had velocities and one had a velocity of 50% stenosis. One limitation of our study is that it is retrospective, however, it is based solely on interpreting data from both sonographic and angiographic images, which were interpreted by different physicians, therefore, randomization may not provide a necessary advantage in analyzing results. There is also some element of variability in imaging, particularly with duplex ultrasound. All efforts were made by each technician to maintain a Doppler angle of 60 degrees or less, so as to allow for minimal variation in this regard. CONCLUSIONS PSV values can be used in detecting 50% and 70% /CA stenosis, and they were better than EDVs and systolic ratios. Previously published data must be validated in individual vascular laboratories. Our results will need prospective validation. AUTHOR CONTRIBUTIONS Conception and design: AA, PS, MS, LD, TK, SH, AM Analysis and interpretation: AA, PS

8 JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 AbuRahma et al 435 Table V. B, Summary of sensitivity, specificity, PPV, NPV, and OA for commonly used criteria for stenosis when applied to our patient population For Sensitivity Specificity PPV NPV OA PSVs 70% stenosis (Moneta) (PSV 275 cm/s) 92% 59% 56% 93% 71% 50% stenosis (Bowersox) (PSV 300 cm/s) 86% 89% 91% 83% 87% EDVs 50% stenosis (Zwolak) (EDV 45 cm/s) 79% 79% 84% 72% 79% 50% stenosis (Perko) (EDV 70 cm/s) 47% 98% 97% 57% 68% NPV, Negative predictive value; OA, overall accuracy; PPV, positive predictive value;, superior mesenteric artery. Table V. C, Summary of sensitivity, specificity, PPV, NPV, and OA for commonly used criteria for CA stenosis when applied to our patient population For CA Sensitivity Specificity PPV NPV OA PSVs 70% stenosis (Moneta) (PSV 200 cm/s) 98% 35% 52% 97% 61% 50% stenosis (Bowersox) (PSV 200 cm/s) 95% 60% 85% 83% 85% EDVs 50% stenosis (Zwolak) (EDV 55 cm/s) 71% 75% 88% 52% 72% 50% stenosis (Perko) (EDV 100 cm/s) 39% 93% 93% 38% 54% EDV, End diastolic velocity; NPV, negative predictive value; OA, overall accuracy; PPV, positive predictive value; PSV, peak systolic velocity. Data collection: MS, TK, SH, AM Writing the article: AA, MS, SH, AM Critical revision of the article: AA, PS, MS, LD, TK, SH, AM Final approval of the article: AA, PS, MS, LD, TK, SH, AM Statistical analysis: AA, LD Obtained funding: Not applicable Overall responsibility: AA REFERENCES 1. Nicholls SC, Kohler TR, Martin RL, Strandness DE Jr. Use of hemodynamic parameters in the diagnosis of mesenteric insufficiency. J Vasc Surg 1986;3: Bowersox JC, Zwolak RM, Walsh DB, Schneider JR, Musson A, LaBombard FE, et al. Duplex ultrasonography in the diagnosis of celiac and mesenteric artery occlusive disease. J Vasc Surg 1991;14: Moneta GL, Yeager RA, Dalman R, Antonovic R, Hall LD, Porter JM. Duplex ultrasound criteria for diagnosis of splanchnic artery stenosis or occlusion. J Vasc Surg 1991;14: Moneta GL, Lee RW, Yeager RA, Taylor LM, Porter JM. Mesenteric duplex scanning: a blinded prospective study. J Vasc Surg 1993;17: Zwolak RM, Fillinger MF, Walsh DB, LaBombard FE, Musson A, Darling CE, et al. Mesenteric and celiac duplex scanning: a validation study. J Vasc Surg 1998;27: Lim HK, Lee WJ, Kim SH, Lee SJ, Choi SH, Park HS, et al. Splanchnic arterial stenosis or occlusion: diagnosis at Doppler ultrasound. Radiology 1999;211: Perko MJ, Just S, Schroeder TV. Importance of diastolic velocities in the detection of celiac and mesenteric artery disease by duplex ultrasound. J Vasc Surg 1997;26: Revzin MV, Pellerito JS. Ultrasound assessment of the mesenteric arteries. Ultrasound Clin 2007;2: DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44: Fillinger MF, Baker RJ Jr, Zwolak RM, Musson A, Lenz JE, Mott J, et al. Carotid duplex criteria for a 60% or greater angiographic stenosis: variation according to equipment. J Vasc Surg 1996;24: Submitted Jan 14, 2011; accepted Aug 10, Additional material for this article may be found online at DISCUSSION Dr Paul Armstrong (Tampa, Fla). Dr AbuRahma and the vascular group from Charleston have endowed the audience with a contemporary overview of mesenteric duplex ultrasonography. Similar to the presentation today, the accompanying manuscript is well prepared. By means of retrospective analysis, the authors identified a group of patients with clinical features of chronic mesenteric ischemia and constructed data analysis using ROC curves to provide validation for duplex interpretation criteria in the vascular laboratory. The target peak systolic and end diastolic values analyzed in this review included both 50% and 70% diameter reducing stenosis of the celiac, superior mesenteric and inferior mesenteric arteries. The information derived from their investigation was also compared with some of the classic work done in the field of visceral duplex ultrasonography. Similar to the Charleston group, our hospital-based vascular laboratories have found peak velocity assessments tend to provide a more accurate account of high-grade mesenteric occlusive disease. This in part may be secondary to the need for clear windows and meticulous sampling technique required to obtain accurate end-diastolic values. Our hospital-based laboratories tend to struggle to secure a good working environment for the intricacies of a pre- and postprandial meal challenge, therefore, we have often

9 436 AbuRahma et al JOURNAL OF VASCULAR SURGERY February 2012 chose to repeat equivocal studies in our office laboratory to be rewarded with a more complete examination. Limitations within the review are chiefly those inherent to a heterogeneous population within the parameters of a retrospective analysis. The authors readily acknowledge these restrictions and cite the additional confines of anatomic variations, which are known to effect visceral duplex velocity patterns. A possible key difference between this report and earlier investigations may be chiefly related to the advancement in our technical understanding of how to perform a proper visceral duplex scan and the vast technologic improvements in image processing and quality assurance over the last several decades. Although not stated in this report, perhaps the greatest utility of mesenteric duplex scanning is not the ability to identify moderate or high-grade stenosis but to exclude the presence of 70% diameter reducing stenosis. Earlier investigations have emphasized the high negative predictive value for visceral duplex, which safely excludes an atherosclerotic contribution to the clinical complaint of chronic mesenteric ischemia. While appraisal of the statistical platforms used in this report provide the reader with an excellent example of the format for quality assurance for vascular laboratory studies, we should be reminded that ROC analysis is useful for validation within a single data subset but provides for poor comparison between multiple groups or series. Thus, the authors have left us with a justifiable conclusion that validation of duplex interpretation criteria should be confirmed within individual vascular laboratories. I enjoyed the manuscript and would like to thank the authors and the association for the opportunity to review their hard work. I have three questions for the authors. (1) In the outpatient setting, we prefer to incorporate a meal challenge especially for an equivocal study. Did your laboratory use a meal challenge as part of your protocol? (2) In this report, you evaluated 85 inferior mesenteric arteries. Are you able to share with the audience any conclusions as to the value of IMA sampling? Did patency or percent stenosis project trends in your velocity criteria used to confirm highgrade occlusive disease? (3) Previous investigations have identified plaque histology, collateral circulation and patterns of vessel stenosis that are clinically important to mesenteric ischemia. In your review of the data, were you able to identify any particular duplex profiles or trends that would aid the interpreter in improving his or her overall accuracy for excluding or predicting clinically significant disease? Dr Ali F. AbuRahma. Thank you, Dr Armstrong, for your comments. We did not use the meal challenge in this group of patients. In regards to the IMA stenosis, unfortunately there were only 15 with 70% stenosis, which made the analysis more difficult, however, we found that a PSV of 270 cm/s was the most accurate for detecting 70% stenosis. Finally, this study only analyzed velocities with no regards to plaque histology or other parameters.

10 JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 AbuRahma et al 436.e1 Appendix Table I (online only). A, Sensitivity, specificity, PPV, NPV, and OA for various velocity cutoffs and ratios for various categories of angiographic 30% stenosis Group No flow or PSV Sensitivity (CI) Specificity (CI) PPV NPV Accuracy (PSV) - 30% stenosis ( ) 94 ( ) ( ) 97 ( ) ( ) 97 ( ) ( ) 97 ( ) ( ) 97 ( ) ( ) 97 ( ) ( ) 97 ( ) (EDV) - 30% stenosis ( ) 18 ( ) ( ) 36 ( ) ( ) 71 ( ) ( ) 82 ( ) ( ) 93 ( ) ( ) 93 ( ) /aortic PSV ratios - 30% stenosis ( ) 3 ( ) ( ) 3 ( ) ( ) 21 ( ) ( ) 42 ( ) ( ) 64 ( ) CA, Celiac artery; EDV, end diastolic velocity; NPV, negative predictive value; OA, overall accuracy; PPV, positive predictive value; PSV, peak systolic velocity;, superior mesenteric artery.

11 436.e2 AbuRahma et al JOURNAL OF VASCULAR SURGERY February 2012 Appendix Table I (online only). B, Sensitivity, specificity, PPV, NPV, and OA for various velocity cutoffs and ratios for various categories of angiographic 50% stenosis Group No flow or PSV Sensitivity (CI) Specificity (CI) PPV NPV Accuracy (PSV) - 50% stenosis ( ) 55 ( ) ( ) 59 ( ) ( ) 66 ( ) ( ) 66 ( ) ( ) 72 ( ) ( ) 75 ( ) ( ) 78 ( ) (81 95) 81 ( ) (81 95) 86 ( ) ( ) 89 ( ) ( ) 89 ( ) ( ) 89 ( ) (64,8 99) 91 ( ) ( ) 97 ( ) ( ) 97 ( ) ( ) 97 ( ) ( ) 97 ( ) ( ) 97 ( ) ( ) 98 ( ) ( ) 100 (100,100) ( ) 100 (100,100) (EDV) - 50% stenosis ( ) 55 ( ) ( ) 64 ( ) ( ) 74 ( ) ( ) 79 ( ) ( ) 83 ( ) ( ) 89 ( ) ( ) 94 ( ) ( ) 98 ( ) ( ) 98 ( ) ( ) 98 ( ) ( ) 98 ( ) /aortic PSV ratios - 50% stenosis ( ) 13 ( ) ( ) 23 ( ) ( ) 45 ( ) ( ) 66 ( ) ( ) 78 ( ) ( ) 86 ( ) ( ) 92 ( ) CA, Celiac artery; EDV, end diastolic velocity; NPV, negative predictive value; OA, overall accuracy; PPV, positive predictive value; PSV, peak systolic velocity;, superior mesenteric artery.

12 JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 AbuRahma et al 436.e3 Appendix Table I (online only). C, Sensitivity, specificity, PPV, NPV, and OA for various velocity cutoffs and ratios for various categories of angiographic 70% stenosis Group No flow or PSV Sensitivity (CI) Specificity (CI) PPV NPV Accuracy (PSV) - 70% stenosis ( ) 59 ( ) ( ) 62 ( ) ( ) 65 ( ) ( ) 65 ( ) ( ) 67 ( ) ( ) 72 ( ) ( ) 76 ( ) ( ) 82 ( ) ( ) 85 ( ) ( ) 86 ( ) ( ) 86 ( ) ( ) 86 ( ) ( ) 87 ( ) ( ) 90 ( ) ( ) 93 ( ) ( ) 95 ( ) (EDV) - 70% stenosis (100,100) 47 ( ) (100,100) 57 ( ) ( ) 66 ( ) ( ) 70 ( ) ( ) 77 ( ) ( ) 81 ( ) ( ) 90 ( ) ( ) 95 ( ) ( ) 95 ( ) ( ) 97 ( ) ( ) 97 ( ) /aortic PSV ratios - 70% stenosis ( ) 56 ( ) ( ) 66 ( ) ( ) 76 ( ) ( ) 83 ( ) ( ) 86 ( ) ( ) 89 ( ) CA, Celiac artery; EDV, end diastolic velocity; NPV, negative predictive value; OA, overall accuracy; PPV, positive predictive value; PSV, peak systolic velocity;, superior mesenteric artery.

13 436.e4 AbuRahma et al JOURNAL OF VASCULAR SURGERY February 2012 Appendix Table II (online only). A, Sensitivity, specificity, PPV, NPV, and OA for various velocity cutoffs and ratios for various categories of angiographic 30% CA stenosis Group No flow or PSV Sensitivity (CI) Specificity (CI) PPV NPV Accuracy Celiac (PSV) - 30% stenosis Celiac ( ) 80 ( ) Celiac ( ) 95 ( ) Celiac ( ) 100 (100,100) Celiac ( ) 100 (100,100) Celiac ( ) 100 (100,100) Celiac ( ) 100 (100,100) Celiac ( ) 100 (100,100) Celiac ( ) 100 (100,100) Celiac (EDV) - 30% stenosis Celiac ( ) 5 ( ) Celiac ( ) 10 ( ) Celiac ( ) 25 (6.0 44) Celiac ( ) 40 ( ) Celiac ( ) 60 ( ) Celiac ( ) 75 ( ) Celiac/aortic PSV ratios - 30% stenosis Celiac ( ) 0 (0,0) Celiac 1 93( ) 10 ( ) Celiac ( ) 20 ( ) Celiac 2 86 ( ) 50 ( ) Celiac ( ) 55 ( ) Celiac ( ) 70 ( ) Celiac ( ) 85 ( ) CA, Celiac artery; EDV, end diastolic velocity; NPV, negative predictive value; OA, overall accuracy; PPV, positive predictive value; PSV, peak systolic velocity.

14 JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 AbuRahma et al 436.e5 Appendix Table II (online only). B, Sensitivity, specificity, PPV, NPV, and OA for various velocity cutoffs and ratios for various categories of angiographic 50% CA stenosis Group No flow or PSV Sensitivity (CI) Specificity (CI) PPV NPV Accuracy Celiac (PSV) - 50% stenosis Celiac ( ) 43 ( ) Celiac ( ) 50 ( ) Celiac ( ) 60 ( ) Celiac ( ) 64 ( ) Celiac ( ) 74 ( ) Celiac ( ) 76 ( ) Celiac ( ) 83 ( ) Celiac ( ) 86 ( ) Celiac ( ) 93 ( ) Celiac ( ) 93 ( ) Celiac ( ) 95 ( ) Celiac (65.4,82.6) 95 ( ) Celiac ( ) 95 ( ) Celiac ( ) 95 ( ) Celiac ( ) 98 ( ) Celiac ( ) 98 ( ) Celiac ( ) 98 ( ) Celiac ( ) 98 ( ) Celiac ( ) 98 ( ) Celiac (EDV) - 50% stenosis Celiac ( ) 20 ( ) Celiac ( ) 35 ( ) Celiac ( ) 48 ( ) Celiac ( ) 58 ( ) Celiac ( ) 70 ( ) Celiac ( ) 75 ( ) Celiac ( ) 85 ( ) Celiac ( ) 88 ( ) Celiac ( ) 93 ( ) Celiac ( ) 93 ( ) Celiac ( ) 93 ( ) Celiac/aortic PSV ratios - 50% stenosis Celiac ( ) 37 ( ) Celiac ( ) 44 ( ) Celiac ( ) 56 ( ) Celiac ( ) 71 ( ) Celiac ( ) 73 ( ) Celiac ( ) 88 ( ) Celiac ( ) 98 ( ) CA, Celiac artery; EDV, end diastolic velocity; NPV, negative predictive value; OA, overall accuracy; PPV, positive predictive value; PSV, peak systolic velocity.

15 436.e6 AbuRahma et al JOURNAL OF VASCULAR SURGERY February 2012 Appendix Table II (online only). C, Sensitivity, specificity, PPV, NPV, and OA for various velocity cutoffs and ratios for various categories of angiographic 70% CA stenosis Group No flow or PSV Sensitivity (CI) Specificity (CI) PPV NPV Accuracy Celiac (PSV) - 70% stenosis Celiac ( ) 24 ( ) Celiac ( ) 28 ( ) Celiac ( ) 30 ( ) Celiac ( ) 35 ( ) Celiac ( ) 39 ( ) Celiac ( ) 45 ( ) Celiac ( ) 48 ( ) Celiac ( ) 52 ( ) Celiac ( ) 57 ( ) Celiac ( ) 61 ( ) Celiac ( ) 64 ( ) Celiac ( ) 65 ( ) Celiac ( ) 67 ( ) Celiac ( ) 70 ( ) Celiac ( ) 71 ( ) Celiac ( ) 82 ( ) Celiac ( ) 89 ( ) Celiac ( ) 90 ( ) Celiac ( ) 90 ( ) Celiac ( ) 93 ( ) Celiac (EDV) - 70% stenosis Celiac ( ) 18 ( ) Celiac ( ) 29 ( ) Celiac ( ) 38 ( ) Celiac ( ) 46 ( ) Celiac ( ) 58 ( ) Celiac ( ) 61 ( ) Celiac ( ) 73 ( ) Celiac ( ) 75 ( ) Celiac ( ) 85 ( ) Celiac ( ) 87 ( ) Celiac ( ) 91 ( ) Celiac/aortic PSV ratio - 70% stenosis Celiac ( ) 53 ( ) Celiac ( ) 72 ( ) Celiac ( ) 83 ( ) Celiac ( ) 87 ( ) Celiac ( ) 89 ( ) Celiac ( ) 95 ( ) Celiac ( ) 95 ( ) CA, Celiac artery; EDV, end diastolic velocity; NPV, negative predictive value; OA, overall accuracy; PPV, positive predictive value; PSV, peak systolic velocity.